Supervised AM–PM milk weighings of Jersey and Holstein cattle from 1966 to 1970 were merged with information from the Canadian Record of Performance cow master file. Factors were derived for predicting daily milk yield from either the morning or evening test samples considering milking interval from PM to AM tests and month of lactation. Morning weighings were found to predict daily yield more accurately than evening weighings. Various schemes for predicting 305-day milk yield were compared. An alternating AM–PM testing scheme was shown to be the best of four schemes for predicting 305-day milk yield, and was only slightly less accurate than using actual daily milk weights.